Int J Primatol
DOI 10.1007/s10764-013-9704-x
A Potential Distribution Model and Conservation
Plan for the Critically Endangered Ecuadorian
Capuchin, Cebus albifrons aequatorialis
Fernando A. Campos & Katharine M. Jack
Received: 28 March 2013 / Accepted: 10 June 2013
# Springer Science+Business Media New York 2013
Abstract Conservation actions that effectively and efficiently target single, highly
threatened species require current data on the species’ geographic distribution and
environmental associations. The Ecuadorian capuchin (Cebus albifrons aequatorialis)
is a critically endangered primate found only in the fragmented forests of western
Ecuador and northern Peru, which are among the world’s most severely threatened
ecosystems. We use the MAXENT species distribution modeling method to model the
potential distribution and environmental associations of Cebus albifrons aequatorialis,
using all known presence localities recorded within the last 2 decades as well as 13
climate, topography, vegetation, and land-use data sets covering the entire geographic
range of the subspecies. The environmental conditions that our model predicted to be
ideal for supporting Cebus albifrons aequatorialis included ≥20% tree cover, mild
temperature seasonality, annual precipitation <2000 mm, and low human population
density. Our model identified 5028 km2 of suitable habitat remaining, although many of
these forest fragments are unprotected and are unlikely to support extant populations.
Using the median population density across all sites for which data are available, we
estimate the total carrying capacity of the remaining habitat to be 12,500 total
individuals. The true number of remaining individuals is likely to be considerably
lower due to anthropogenic factors. We highlight four critical regions of high
predicted suitability in western Ecuador and northern Peru on which immediate
conservation actions should focus, and we lay out clear priorities to guide
Electronic supplementary material The online version of this article (doi:10.1007/s10764-013-9704-x)
contains supplementary material, which is available to authorized users.
F. A. Campos (*)
Department of Anthropology, University of Calgary, Alberta T2N 1N4, Canada
e-mail: [email protected]
F. A. Campos
Department of Integrated Biosciences, University of Tokyo, Kashiwa, Chiba 277-8562, Japan
K. M. Jack
Department of Anthropology, Tulane University, New Orleans, Louisiana 70118, USA
F.A. Campos, K.M. Jack
conservation actions for ensuring the long-term survival of this gravely threatened and
little known primate.
Keywords Cebus albifrons aequatorialis . Conservation plan . Distribution model .
Ecuadorian capuchin . MAXENT . Northern Peru . Western Ecuador
Introduction
As the funds and resources available for conservation are scarce, decisions that focus
on conserving single, highly threatened species must have a sound basis, including
knowledge of the species’ current geographic distribution and environmental associations. Although such “focal-species” approaches to conservation have been criticized (Lindenmayer et al. 2002), certain species make logical conservation targets
because they may be successful at raising public awareness and increasing available
conservation funds (Lambeck 1997; Wilson et al. 2009). The tropical forests of
western Ecuador and northern Peru are among the most imperiled ecosystems on
earth (Dodson and Gentry 1991), yet they constitute the only refuge of the critically
endangered Ecuadorian capuchin, Cebus albifrons aequatorialis (Allen 1914;
Cornejo and de la Torre 2008). Dodson and Gentry (1991) suggest that only 4.4%
of the original forest remains in western Ecuador, and many of the remaining forest
tracts in this region have been severely degraded by human disturbance that is largely
attributable to exponential human population growth in the region since 1960. This
rapid habitat loss has had devastating consequences on the region’s unique biodiversity (Brooks et al. 2002; Dodson and Gentry 1991; Parker and Carr 1992), leading to
Conservation International’s designation of the region as one of the world’s leading
biodiversity hotspots (Myers et al. 2000). This rampant deforestation also played a
key role in the recent decision by the International Union for Conservation of Nature
(IUCN) to revise the conservation status of Cebus albifrons aequatorialis from near
threatened to critically endangered (Cornejo and de la Torre 2008; Tirira 2011).
Capuchins in western Ecuador and northern Peru are likely to be good indicators of
overall ecosystem health because they are conspicuous, require relatively large areas
of forest (Jack and Campos 2012), and tend to disappear from small, isolated forest
fragments and heavily disturbed areas (Bierregaard 2001; Peres 2001). Thus, Cebus
albifrons aequatorialis can serve as umbrella species for identifying and delineating
areas of high-quality forest that may support more cryptic species that are also in
danger of extinction (Caro and O'Doherty 1999). Moreover, primates in the
Neotropics have been successfully used as flagship species to attract public attention
and support for conservation initiatives (Caro and O'Doherty 1999; Kleiman and
Mallinson 1998).
Extant populations of Cebus albifrons aequatorialis have been recorded in the last
2 decades at a total of 20 localities in western Ecuador and northern Peru (Albuja and
Arcos 2007; Charlat et al. 2000; Cornejo and de la Torre 2008; Encarnacion and
Cook 1998; Hores 2006; Jack and Campos 2012; Parker and Carr 1992). Here, we
model the suitability of habitat for Cebus albifrons aequatorialis across its geographic range by combining published locality data on the species’ occurrence with
remotely sensed environmental data. We provide estimates of the total carrying
Distribution Model and Conservation Plan for Ecuadorian Capuchin
capacity of remaining suitable habitat as well as the current total population size. We
use the habitat suitability model to determine which environmental variables most
strongly predict the species’ occurrence. We also provide a predictive map of the
species’ potential distribution that can be used to identify additional areas where
Cebus albifrons aequatorialis might occur, and to guide conservation actions for C. a.
aequatorialis and other threatened organisms in this ecosystem. Finally, we identify
four critical areas on which immediate conservation actions should focus, and we
suggest specific conservation priorities that will help to ensure the continued survival
of this gravely threatened and little known primate.
Methods
Locality Data
The presence/absence data include 42 sites in western Ecuador and extreme northern
Peru. We carried out surveys for the presence of Cebus albifrons aequatorialis at 11
of these sites between 2002 and 2005 (Jack and Campos 2012). The other 31
localities are drawn from published sources, including a rapid assessment survey
undertaken by a team from Conservation International (Parker and Carr 1992) and the
published reports of Albuja (2002), Albuja and Arcos (2007), Charlat et al. (2000),
Encarnacion and Cook (1998), Gavilanez-Endara (2006), and Hores (2006) (Table I;
Fig. 1). We used the online databases provided by DarwinNet (www.darwinnet.org)
and BirdLife International (www.birdlife.org) and the Google Earth application to
verify that the coordinates provided for each site corresponded to the areas described.
It was evident that a small number of the published coordinates corresponded to
populated areas or administrative areas at the periphery of the surveyed sites, which
might poorly reflect the actual areas used by Cebus albifrons aequatorialis at the
scale of our habitat suitability model. Therefore, we used satellite imagery in Google
Earth to adjust the coordinates for these sites to a forested point at the approximate
geographic center of the protected area. Although this introduces a bias in the model
toward forested areas, we feel that it is an appropriate adjustment based on knowledge
of this arboreal primate’s behavioral ecology. We excluded the presence locality
Cerros de Amotape National Park, Peru, which is reported to have Cebus albifrons
aequatorialis (Cornejo and de la Torre 2008), from our model because we were
unable to find geographic coordinates for a confirmed sighting locality. As this
national park encompasses 913 km2 of variable habitat, designating a single pixel
as the presence locality would have been arbitrary.
Modeling Habitat Suitability
We modeled the habitat suitability of Cebus albifrons aequatorialis with MAXENT
version 3.3.3k (Phillips et al. 2006). Although many methods are available for
modeling species’ distributions, comparative studies have found that MAXENT exhibits
high predictive accuracy for a wide range of species in diverse regions, even with
small sample sizes (Elith et al. 2006; Hernandez et al. 2006; Pearson et al. 2007;
Vidal-García and Serio-Silva 2011). Moreover, our absence data do not fully
F.A. Campos, K.M. Jack
Table I Locality data used to develop the habitat suitability model for Cebus albifrons aequatorialis
Site
Latitude
Longitude
C. a. aequatorialis
present?
Map
number
Source
Bilsa Biological Reserve
0.37656
–79.71080
Present
9
1
Cerro Azul
–3.46842
–79.73256
Present
34
2
Cerro Blanco Protected Forest
–2.15775
–80.04076
Present
23
3, 4
Chirije
–0.70270
–80.47306
Present
12
4
Cooperativa 31 de Agosto
–3.15884
–79.71623
Present
30
2
Cerro de Hayas-Naranjal Reserve
–2.79960
–79.65842
Present
27
2
Cordillera La Tapada
–2.00811
–80.39858
Present
22
2
El Palmar
–0.02684
–80.11596
Present
4
2
Hacienda el Paraíso
–0.26006
–80.35264
Present
7
4
Jauneche
–1.24866
–79.65691
Present
17
3, 4
La Hesperia Biological Reserve
–0.36652
–78.86958
Present
10
4
Machalilla N.P. / La Mocora
–1.61419
–80.71027
Present
19
2, 3, 5
La Planada
–1.33370
–80.65165
Present
18
2
Lalo Loor Dry Forest Reserve
–0.09510
–80.14211
Present
5
4
Loma Alta Ecological Reserve
–1.85520
–80.61131
Present
12
2, 4
Manglares-Churute Ecological Reserve
–2.41250
–79.63384
Present
15
2, 4
Cerro Pata de Pájaro Protected Forest
0.02055
–79.97897
Present
3
3
Tito Santos Biological Reserve
–0.14481
–80.20775
Present
6
2, 4
Tumbes Reserved Zone
–3.82205
–80.25665
Present
39
6
Bosque Petr. de Puyango
–3.85200
–80.02724
Absent
40
2
Cabo Pasado
–0.39605
–80.47390
Absent
8
4
Casacay (Río Jubones)
–3.32606
–79.71869
Absent
33
2
Destacament Cap. Díaz
–3.49592
–80.15229
Absent
35
2
El Guayacán (La Maná)
–0.85950
–79.12073
Absent
13
2
Jaramijó
–0.99589
–80.58914
Absent
14
2
La Comuna (Soledad)
–3.22123
–79.65188
Absent
31
2
La Cotona
0.55232
–79.97143
Absent
1
2
La Liga de Oro
–3.09167
–79.70900
Absent
29
2
Las Planchas
–1.77011
–79.23085
Absent
20
2
Las Vegas
–4.07994
–80.40121
Absent
42
2
Mangaurquillo
–4.00820
–80.19261
Absent
41
2
Manta Real
–2.56697
–79.34944
Absent
25
3
Montecristi
–1.05894
–80.66233
Absent
15
2
Palo Marcado
–3.25865
–79.68515
Absent
32
2
Ramón Campaña
–1.13297
–79.07910
Absent
16
2
Cerro Seco Biological Reserve
–0.61103
–80.43605
Absent
11
4
Buenaventura Reserve
–3.65181
–79.74549
Absent
37
2
Arenillas Ecological Reserve
–3.51681
–80.13337
Absent
36
3
San Isidro
–0.36667
–80.18334
Absent
9
4
San Pedro (Eloy Alfaro)
–2.86388
–79.61401
Absent
28
2
Vía Balsas Marcabelí
–3.76392
–79.86790
Absent
38
2
Distribution Model and Conservation Plan for Ecuadorian Capuchin
Table I (continued)
Site
Latitude
Longitude
C. a. aequatorialis
present?
Map
number
Source
Vía Jesús Maria
–2.65888
–79.44700
Absent
26
2
Sources: 1, Charlat et al. (2000); 2, Albuja and Arcos (2007); 3, Parker and Carr (1992); 4, Jack and
Campos (2012); 5, Hores (2006); 6, Encarnacion and Cook (1998).
represent the range of habitats from which Cebus albifrons aequatorialis is absent;
we therefore chose a modeling method that uses presence data only. We constructed
the model using all 19 known presence localities for Cebus albifrons aequatorialis
(Table I). MAXENT uses occurrence localities to model environmental suitability based
on a set of environmental variables that are likely to influence the species’
Fig. 1 Presence localities (white circles) used to construct the potential distribution model for Cebus
albifrons aequatorialis, and absence localities (black circles) used to evaluate the model. Locality names
and references are listed in Table I. The shaded region represents the maximum geographic extent of C. a.
aequatorialis. We excluded all areas outside this region for constructing the potential distribution model.
F.A. Campos, K.M. Jack
occurrence. MAXENT calculates a value of relative suitability for each cell in the study
area based on the relationship between the presence localities and the environmental
variables measured at each cell. We used default values for convergence threshold
(10–5), maximum number of iterations (500), regularization multiplier (1), and
maximum number of background points (104). We present the model’s output in
logistic format, which assigns to each cell a probability of presence between 0 and 1.
Although MAXENT output is often interpreted as the probability of a species’
occurrence, additional factors not included in our model, such as hunting,
geographic barriers, history, and biotic interactions, are likely to influence the
occupancy of suitable environments (Robinson et al. 2010). Therefore, we refer the
model’s output as a map of potential distribution, with values indicating habitat
suitability rather than probability of occurrence.
Environmental Variables
We compiled 13 climate, topography, vegetation, and land-cover data sets (Table II)
in ESRI ArcGIS 9.3.1 as input variables in the habitat suitability model. Each
environmental data set was a grid with identical geographic limits and cell size
covering the known geographic range of Cebus albifrons aequatorialis in western
Ecuador and northern Peru (Fig. 2). All data sets were converted to the WGS 1984
geographic coordinate system. We attempted to select data sets that were collected as
close as possible to the 2002–2005 time period, during which most of the
presence/absence survey localities were recorded. The 13 environmental variables
are numbered and described in the text that follows.
We assessed 19 “bioclimatic variables” from WorldClim 1.4 (Hijmans et al. 2005;
http://www.worldclim.org/) for strength of pairwise correlations and discarded one
variable at random from highly correlated pairs (|r| > 0.85). This selection process
resulted in six bioclimatic variables used for the model: 1) mean diurnal temperature
range (°C), 2) temperature seasonality (°C), 3) minimum temperature of the coldest
month (°C), 4) temperature annual range (°C), 5) annual precipitation (mm), and 6)
precipitation seasonality (coefficient of variation). These data sets are derived from
interpolated mean monthly temperature and rainfall data collected at weather stations
Table II Summary of contributions by the 13 environmental variables to the habitat suitability model
1
MeanDiTempRange
Mean diurnal temperature range (° C)
7 .2 0
Training
gain
without
1 .5 7
0 .1 2 0
0 .0 0
2
TempSeasonality
Temperature seasonality (°C)
0.011
1.57
0.133
0.126
3
MinTempColdestM
Minimum temperature in coldest month (°C)
1.48
1.53
0.0448
5.56
4
TempAnnualRange
Temperature annual range (°C)
2.52
1.52
0.254
28.8
5
AnnualPrecip
Annual precipitation (mm)
1 1 .2
1.36
0.125
20.9
6
PrecipSeasonality
Precipitation seasonality (coefficient of variation)
0.00
1.57
0.0291
0.00
7
Elevation
Elevation (m)
3.54
1.55
0.218
4.79
Name
Description
Percent
contribution
Training gain
with only
Permutation
importance
8
Slope
Slope (degrees to the horizontal)
6 .4 1
1.54
0.309
2.87
9
TreeCover
Percent tree cover
36.9
1.43
0.651
18.0
10
NDVIMean
Mean of normalized difference vegetation index (NDVI)
0.113
1.57
0.165
0.00
11
NDVIStdDev
Standard deviation of NDVI
1 .5 4
1.55
0.0752
0.600
12
HumanPopDensity
Square root of human population density (per km2)
12.8
1.51
0.405
0.0067
13
LandCover
Land cover category
16.3
1.35
0.467
18.3
Boxed cells indicate values that are at least one standard deviation from the mean across all variables for the
given measure of variable importance.
Distribution Model and Conservation Plan for Ecuadorian Capuchin
MeanDiTempRange, °C TempSeasonality, °C
12
10
TempAnnualRange, °C
22
18
20
20
16
18
14
16
12
14
10
12
8
15
8
10
6
5
4
0
AnnualPrecip, mm
MinTempColdestM, °C
25
PrecipSeasonality, CV
Slope, ° to the horizontal
Elevation, m
4500
220
1600
4000
200
1400
3500
180
1200
25
3000
160
1000
20
2500
140
800
2000
120
600
100
400
1500
1000
80
500
60
0
NDVIMean
80
70
5
0
0.8
50
0.4
40
10
0
HumanPopDensity1/2, per km2
0.6
0.6
20
10
0
NDVIStdDev
60
30
15
200
40
TreeCover, %
30
0.5
200
0.4
150
0.3
100
0.2
0.2
0.0
0.1
50
−0.2
0.0
0
Fig. 2 Maps of the 12 continuous environmental data sets used to construct the potential distribution
model for Cebus albifrons aequatorialis. An additional categorical data set used to construct the model,
land cover, is shown in Online Resource 1. Variable names are described in Table II.
from ca. 1950 to 2000. We also acquired from WorldClim 1.4 a digital elevation
model (DEM) to represent 7) elevation (m) and 8) slope (degrees to the horizontal).
We derived slope from the DEM using the ArcInfo Spatial Analyst extension. We
used 9) the Vegetation Continuous Fields Tree Cover product (M. Hansen et al. 2007;
Global Land Cover Facility, www.landcover.org), which depicts proportion of tree
canopy cover for each pixel based on light penetration to the ground. This data set is
provided at a spatial resolution of 500 × 500 m (15.1632' ') at the equator. To cover
the extent of the study area, we mosaicked two coverages from 2005. We used
MODIS Terra Vegetation Indices (Global MOD13A3 data) to derive two variables:
10) annual mean Normalized Difference Vegetation Index (NDVI) and 11) standard
deviation of annual NDVI for the year 2004. NDVI is a widely used measure of green
vegetation based on the wavelengths of radiation characteristically reflected and
absorbed by chlorophyll; therefore, NDVI values correlate strongly with the photosynthetic capacity of vegetation (Myneni et al. 1995). Global MOD13A3 data are
provided monthly at 1-km spatial resolution and are distributed by the Land Processes
Distributed Active Archive Center (LP DAAC), located at the U.S. Geological
Survey (USGS) Earth Resources Observation and Science (EROS) Center
F.A. Campos, K.M. Jack
(lpdaac.usgs.gov). We used the Raster Calculator in the ArcInfo Spatial Analyst
extension to calculate mean NDVI and standard deviation of NDVI based on four
monthly NDVI layers from 2004 corresponding to February (peak wet season), May
(wet–dry transition), August (peak dry season), and November (dry-wet transition). We
used calculated 12) human population density from population count and area grid data
from the LandScan 2008 High Resolution global Population Data Set. Because the raw
population density values exhibited extremely high variance, we applied a square root
transform. Finally, we used 13) the Vegetation Map of Latin America (Eva et al. 2002), a
thematically detailed land-cover map of South America produced for the Global Land
Cover 2000 database coordinated by the European Commission's Joint Research Centre
(Bartholome and Belward 2005; GLC 2003). This data set features 73 regionally
optimized land-cover classes based on differences in vegetation, geology, elevation,
rainfall, and anthropogenic disturbance. We reclassified the 73 regional classes into 15
global classes using the global legend transformation in Fritz et al. (2003).
We clipped all rasters to exclude areas outside the known range of Cebus albifrons
aequatorialis, with boundaries as follows: to the west the Pacific Ocean; to the north
the Esmeraldas-Guayllabamba River (Jack and Campos 2012); to the east a contour
line along the Andes mountains at 1500 m, the maximum recorded elevation for the
subspecies (Allen 1914; Harris et al. 2008); and to the south the latitude –4.66, which
is ca. 100 km south of the southernmost presence locality and south of which the
landscape appears too arid to support Cebus populations (Fig. 1). For all data sets, we
adjusted grid cell resolution and alignment to match the Worldclim 1.4 data, which
have a resolution of 30' ' (0.93 × 0.93 km at the equator).
Model Evaluation
Presence-only species distribution models are typically validated with an independent
set of presence points by assessing how accurately the model predicts occurrence at
these points. However, Pearson et al. (2007) note that when the models are based on a
small number of localities, the output can be strongly influenced by exactly which
localities are included. Therefore, we used all available presence points as inputs in
the model and used two alternative methods to assess the model’s performance.
First, we assessed the area under the receiver operating characteristic curve (Manel
et al. 2001), which measures the model’s ability to correctly distinguish presence localities
from randomly chosen background pixels (Manel et al. 2001; Phillips et al. 2006). AUC
values of >0.9 indicate high accuracy. Second, we used the threshold-based jackknife
validation procedure described by Pearson et al. (2007). Many practical applications of a
species distribution model require that the model’s continuous, probabilistic output be
converted to a binary map (suitable/unsuitable). This conversion involves specifying a
threshold of occurrence, with suitable conditions predicted above the threshold and
unsuitable conditions predicted in the text that follows. There are a variety of methods
for selecting threshold values (Liu et al. 2005), and the best strategy may vary with
different research and management objectives (Jimenez-Valverde and Lobo 2007;
Loiselle et al. 2003; Robinson et al. 2010). In particular, the potential costs of omission
errors (predicting absence at a true presence site) must be weighed against the potential
costs of commission errors (predicting presence at a true absence site) for the system
under study. For conservation applications with rare and highly endangered species, there
Distribution Model and Conservation Plan for Ecuadorian Capuchin
is a delicate balance to be struck: too-conservative a threshold (underestimation) may
cause failure to correctly identify new populations in need of protection (Robinson et al.
2010), whereas too-lenient a threshold (overestimation) may cause misdirection of critical
conservation actions (Loiselle et al. 2003). In such cases, particularly with small sample
sizes and incomplete absence data, Pearson et al. (2007) recommend using the “lowest
presence threshold”: the lowest predicted value for habitat suitability at any observed
presence locality. However, the locality with the lowest presence locality for our data set
was a clear outlier (see Results), suggesting that the lowest presence threshold might
overestimate considerably the potential distribution of Cebus albifrons aequatorialis.
Therefore, we examined several other commonly used candidate threshold values. Two
recent studies that evaluated the performance of a variety of threshold-selection methods
agree that the method introduced by Manel et al. (2001) performs well (Jimenez-Valverde
and Lobo 2007; Liu et al. 2005). This threshold corresponds to the value at which
discrimination accuracy is maximized between true presences and pseudo-absences
(random background points). Specifically, it uses the point on the receiver operating
characteristic curve at which the sum of sensitivity and specificity for the training data is
maximized (Manel et al. 2001). Two additional lines of evidence suggest that this
threshold value is appropriate for our data set. First, it happens to be equal to the next
lowest presence threshold if we were to exclude the one extreme outlier, as well as the
commonly used 10-percentile presence threshold. Second, this value nearly minimized
the number of omission and commission errors among the survey locality data (presence
localities below the threshold and absence localities above the threshold).
Suitable Habitat Assessment and Population Estimates
Using the maximum training sensitivity plus specificity threshold (Manel et al. 2001), we
calculated the total area of habitat predicted to be suitable for Cebus albifrons
aequatorialis across its entire geographic range. This resulted in many small, isolated
pixels (area 0.93 km2) that are unlikely to be capable of supporting viable populations of
this primate. To remove these pixels, we applied a majority filter to the suitable habitat
map (Fig. 3). This process replaced suitable/unsuitable cells if at least half of the four
neighboring orthogonal cells were predicted to be of the other suitability category. We also
Fig. 3 Illustration of applying the habitat suitability threshold and majority filter method used to discriminate suitable from unsuitable habitat, remove isolated pixels, and clean the boundaries of the areas
predicted to be suitable for Cebus albifrons aequatorialis. The left panel shows raw output from the
Maxent algorithm, with darker shading representing more suitable habitat. The middle panel shows the
resulting habitat estimate after applying the suitability threshold; all areas with predicted suitability below
the threshold are excluded. The right panel shows the final habitat estimate after applying the majority filter.
F.A. Campos, K.M. Jack
calculated a coarse estimate of the remaining total population size by multiplying the
median population density across all sites for which data were available (2.4
individuals/km2; Jack and Campos 2012) by the area of remaining suitable habitat. The
median population density reported by Jack and Campos (2012) excludes the presence
locality Jauneche, which was a clear outlier, with very low estimated suitability in this
study (see Results) but extraordinarily high population density, perhaps owing to the
abundant crops on which the monkeys feed (Jack and Campos 2012).
Results
Model Predictions and Performance
The area under the receiver operating characteristic curve for the habitat suitability
model was 0.971, which indicates that the model was highly effective at discriminating presence localities from random background pixels. The predicted suitability
values for all 42 survey localities are plotted in Fig. 4. The model was reasonably
successful at discriminating the 19 presence localities from the 23 absence localities,
which were not used to generate the model (18/19 presence localities and 19/23
absence localities categorized correctly), despite the fact that the absence localities
represent forested areas, some of which are under protection. The potential
1.00
Predicted Suitability
0.75
0.50
Suitability Threshold
0.25
0.00
Absence
Presence
Sites
Fig. 4 Violin plots of predicted habitat suitability for Cebus albifrons aequatorialis at presence and
absence localities. Violin plots combine elements of standard box plots with kernel density plots, and are
useful for visualizing dispersion and skew in the data. Black circles represent the 42 localities, and the white
boxplots show median values as well as first and third quartiles. The dashed horizontal line represents the
threshold value used to discriminate suitable from unsuitable habitat.
Distribution Model and Conservation Plan for Ecuadorian Capuchin
distribution map identified four general regions containing substantial patches of high
predicted suitability: the Chongon-Colonche hills near the coast of south-central
Ecuador and west of Guayaquil, the northern coast of the Manabí province in
Ecuador, the foothills of the Andes Mountains in southern Ecuador, and the
Tumbes and Piura regions of northern Peru (Fig. 5).
Environmental Factors and Habitat Suitability
Of the 13 environmental variables investigated, 4 provided important contributions to
the habitat suitability model (Table II). The MAXENT jackknife analysis revealed that
the most important variable for determining habitat suitability was percent tree cover.
Inspection of the individual variables’ response curves —which show how predicted
suitability changes as each environmental variable changed while keeping the other
variables at their mean values— indicated that habitat suitability is maximized when
percent tree cover is ≥25%, above which suitability appears to plateau (Fig. 6). The
Second Priority Region:
Mache-Chindul and Jama-Coaque
mountain ranges
First Priority Region:
Chongon-Colonche
mountain range
Fourth Priority Region:
Andean foothills of
southern Ecuador
Third Priority Region:
North-West Peru Biosphere Reserve
Fig. 5 Map of suitable habitat (red shaded areas) for Cebus albifrons aequatorialis across its entire
geographic range. Suitable areas contained within the four indicated priority regions (bright red with a
yellow outline) are of particular conservation concern.
F.A. Campos, K.M. Jack
Predicted Suitability
TempAnnualRange
AnnualPrecip
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
7.5
10.0
12.5
15.0
17.5
0
1000
TreeCover
2000
3000
4000
HumanPopDensity
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0
25
50
75
0
50
100
150
200
Value
Fig. 6 Response curves for four variables found to be important for determining habitat suitability for
Cebus albifrons aequatorialis: temperature annual range (°C), annual precipitation (mm), percent tree cover
(%), and square root transformed human population density.
importance attributed to percent tree cover was evident in all four measures of
variable importance (Table II). In accordance with this finding, land-cover category
was identified as important in three of the measures of variable importance (Table II),
with the most suitable habitat occurring in deciduous or evergreen broadleaved forest,
or to a lesser degree, forested cropland mosaic. Annual precipitation, temperature
annual range, and human population density appeared to be relatively important in
some tests but not others. These discrepancies commonly occur with the MAXENT
algorithm owing to correlations among the predictor variables, but nonetheless
suitable habitat tended to exhibit relatively low annual precipitation, low annual
temperature range, and low human population density (Fig. 6). Percent tree cover
and land cover exhibited the highest gain when used in isolation to build the model
(Table II), which indicates that these two variables contained the most relevant
information for predicting habitat suitability. The removal of percent tree cover,
annual precipitation, and land cover caused the greatest reduction in gain during
the jackknife procedure (Table II), which indicates that these three variables
contained the most information that was absent from the other variables.
Habitat Suitability and Maximum Population Size
The habitat suitability threshold that maximized discrimination accuracy was 0.381.
Applying this threshold to the model results produced an estimate of 5208 km2 of
suitable habitat (shaded area in Fig. 5) and a total carrying capacity of 12,500
individuals if all habitats were occupied at the median population density. The
jackknife validation procedure indicated that this threshold-based scenario was significantly better than random at predicting the presence of Cebus albifrons
Distribution Model and Conservation Plan for Ecuadorian Capuchin
aequatorialis in the study region (prediction success 12/19, P ≈ 0). The apparently
modest success among jackknife iterations at correctly predicting the excluded
locality as a presence point indicates that the specific localities included in each
model influenced the resulting habitat estimates considerably. This is likely due to our
small sample size, which further justifies our decision to use all presence points for
the final model. The very small P-value despite this modest success rate is due to the
fact that only a small proportion of the study area was considered suitable (mean
0.0753, range 0.0400–0.0964).
Discussion
Habitat Associations of the Ecuadorian Capuchin
Although extant populations of Cebus albifrons aequatorialis occur in a range of
habitats including dry forest, mature moist forest, and premontane forest (Albuja and
Arcos 2007; Jack and Campos 2012; Parker and Carr 1992), our results suggest that
their continued survival may well depend on conservation of the acutely threatened
Ecuadorian and Tumbes-Piura tropical dry forests with which the species appears
most closely associated. The environmental conditions that our model predicted to be
ideal for supporting Cebus albifrons aequatorialis —including ≥25% tree cover, mild
temperature seasonality, annual precipitation below 2000 mm, and low human
population density— more accurately describe tropical dry forest than other forest
biomes in this region. A threshold value of 25% tree cover has been used previously
with MODIS VCF data to discriminate forest from nonforest pixels (Liknes et al.
2010). Our finding that habitat suitability reaches a plateau between 25% and 100%
tree cover is consistent with our previous observations that Cebus albifrons
aequatorialis readily exploits a variety of habitat types ranging from disturbed areas
and secondary forest to mature forest. Pixels with tree cover values at the lower end
of the suitable range (25–60%) are likely to represent relatively open woodlands and
secondary forest (M. C. Hansen et al. 2000). As in a recent study that identified areas
in Mexico likely to be inhabited by primates (Vidal-García and Serio-Silva 2011), the
results of our modeling approach facilitate the difficult task of establishing clear
conservation priorities by revealing unsurveyed areas that may harbor undetected
populations and identifying the areas that may be most capable of supporting viable
Cebus albifrons aequatorialis populations in the long term.
The total current extent of Ecuadorian and Tumbes-Piura dry forest in western
Ecuador and northern Peru was estimated recently to be 7940 km2 (Portillo-Quintero
and Sánchez-Azofeifa 2010), which is comparable to, but somewhat larger than, the
5028 km2 of suitable habitat estimated by our potential distribution map. A visual
comparison between our habitat estimate and the tropical dry forest map in PortilloQuintero and Sánchez-Azofeifa (2010) indicates that our habitat estimate excludes
sizable areas of relatively sparse dry forest that were included in Portillo-Quintero and
Sánchez-Azofeifa’s estimate, especially in central Manabí and northern Guayas
provinces western Ecuador. However, our estimate includes additional regions humid
premontane and lower montane cloud forest in the Andean foothills of central and
southern Ecuador.
F.A. Campos, K.M. Jack
A Realistic Population Size Assessment
Although we have estimated the total carrying capacity of the remaining suitable
habitat to be 12,500 total individuals, we make the case below that the true total
population size of Cebus albifrons aequatorialis is likely to be considerably smaller.
First, the carrying capacity figure assumes that all suitable habitats are occupied. This
assumption is certainly false, as extant populations of Cebus albifrons aequatorialis
are known to be absent from some highly suitable areas that have been extensively
surveyed (Fig. 4), and they have no doubt been extirpated from other unsurveyed
areas that may still appear to be suitable (one example is described in Jack and
Campos 2012). It is possible that some extant Cebus albifrons aequatorialis populations occur in areas predicted to be unsuitable in our model; Jauneche (Fig. 1, locality
17) is one such locality. However, we believe that these omission errors are probably
rare and are unlikely to offset the more common commission errors described above.
Second, the carrying capacity figure assumes that all occupied habitats have a
population density equal to the median value from sites with available data (Albuja
and Arcos 2007; Jack and Campos 2012). However, four of the five sites from which
population density was calculated have some form of protected status, whereas the
vast majority of suitable habitat identified in our model is under no official protection.
We believe it is likely that population densities of Cebus albifrons aequatorialis will
be higher in protected forests than in unprotected forests owing to the combined
effects of hunting pressure, tree felling, uncontrolled fires associated with land
clearing, and other anthropogenic factors. Third, the highly fragmented configuration
of suitable habitat remaining suggests that occupied patches will have suffered
increased probability of local extinctions, with little chance of recolonization after a
local extinction (Brooks et al. 2002). We currently lack sufficient data to state with
confidence how much the carrying capacity figure that we calculated overestimates
true population size. However, given the issues discussed in the preceding text, we
believe it is not unreasonable to speculate that the true total population size may be
less than half of the carrying capacity, which suggests a total population size of fewer
than 6250 individuals.
Focal Regions and Conservation Priorities
We highlight four key regions, in descending order of importance, that we deem
critical for the long-term survival of Cebus albifrons aequatorialis in western
Ecuador and northern Peru (Fig. 5). These priority regions were not ranked based
on a quantitative scheme for optimal resource allocation. Such an analysis —which
would require a comprehensive understanding of monetary costs, the likelihood of
success, and the social and political feasibility for each proposed action— is beyond
the scope of this study (Game et al. 2013; Wilson et al. 2006, 2009). Given the recent
history of catastrophic forest loss across western Ecuador (Dodson and Gentry 1991),
all of the remaining areas containing suitable habitat for Cebus albifrons
aequatorialis should be considered highly threatened and irreplaceable. Our ultimate
goal for setting conservation priorities here is to maximize the number of Cebus
albifrons aequatorialis individuals remaining across the region. Thus, we give greater
priority to areas with 1) greater extent of continuous and/or undisturbed forest and 2)
Distribution Model and Conservation Plan for Ecuadorian Capuchin
greater likelihood of conservation success due to existing (though possibly inadequate) protection.
The area of highest priority is located along the Chongón-Colonche range of
coastal mountains in the Guayas and Manabí provinces of Ecuador (Fig. 5, First
Priority Region). We believe that conservation efforts aimed at improving protection
in the Chongon-Colonche range, including Machalilla National Park (Fig. 1, locality
19), could have the most significant impact on this primate’s long-term survival
prospects. There are three main reasons for this: 1) the Chongon-Colonche range
represents the largest and most continuous region of high predicted suitability across
the entire geographic range of Cebus albifrons aequatorialis; 2) although officially
protected, Machalilla National Park suffers from intense human pressures that have
led to habitat degradation in many areas of the park (Best and Kessler 1995; Parker
and Carr 1992); and 3) between Machalilla National Park and Cerro Blanco (Fig. 1,
locality 23), there is a large area that currently has no official protection aimed at
preserving biodiversity. South of Machalilla National Park, the tropical dry forest
covering the upper parts of the Chongon-Colonche range has been managed since
1994 to maintain its important function as a rain catchment area, but the wildlife is
not protected (Best and Kessler 1995). Improving the protection of biodiversity in the
Chongon-Colonche range could have long-term ancillary benefits for human inhabitants in the region, as the preservation of pollinating and seed-dispersing animals
may be necessary for the forested hills to retain their economic value as a rain
catchment. Finally, the Chongon-Colonche range has great potential as an ecotourism
destination, given its unique flora and avifauna, which includes many charismatic
range-restricted or endemic species (Best and Kessler 1995; Parker and Carr 1992).
The largest continuous block of suitable habitat in this region extends ca. 80 km from
Machalilla National Park in the northwest to the town of Simón Bolívar in the
southeast. About 20 km north of this area, near Cerro Achi (La Planada, locality 18
in Fig. 1), there is another small area of high predicted suitability that is separated
from the former area by gaps of low suitability within and just outside of Machalilla
National Park. There are at least four localities in this region inhabited by Cebus
albifrons aequatorialis (Albuja and Arcos 2007; Hores 2006; Jack and Campos 2012;
Parker and Carr 1992). A fifth inhabited site, the Cerro Blanco Protected Forest
(Fig. 1, locality 23), is separated from the main block of suitable habitat in the
Chongon-Colonche range by ca. 20 km of unsuitable habitat.
The second area of high priority is a chain of small suitable areas in the MacheChindul and Jama-Coaque mountain ranges, which extend north from Bahía de
Caráquez to Mache-Chindul National Park (Fig. 1, locality 2) along the northern
coast of Ecuador’s Manabí Province and southern Esmeraldas Province (Fig. 5,
Second Priority Region). The suitable areas in Manabí appear highly discontinuous,
with many relatively small and isolated forest fragments separated by gaps of
unsuitable habitat. Surveys revealed the presence of Cebus albifrons aequatorialis
in at least six of these fragments, two of which are privately owned, protected areas:
the Lalo Loor Dry Forest Reserve (Fig. 1, locality 5) and the Tito Santos Biological
Reserve (Fig. 1, locality 6) (Albuja and Arcos 2007; Jack and Campos 2012). Forest
fragments in this region are gravely threatened by logging, cattle, and human
settlement. A recent local extinction of Cebus albifrons aequatorialis has been
recorded in at least one such forest fragment (Jack and Campos 2012). The largest
F.A. Campos, K.M. Jack
tract of suitable habitat in this region lies within and around the southern portion of
Mache-Chindul National Park. An extant population of Cebus albifrons aequatorialis
is known to occur here, and although the park remains poorly surveyed, it may harbor
biodiversity of potentially great conservation importance (Best and Kessler 1995). An
effective conservation plan for this region should focus on shoring up existing
protection by supporting the efforts of private reserve owners, promoting connectivity
among remaining forest fragments via reforestation along biological corridors, and
determining the conservation value of Mache-Chindul National Park.
The third priority area includes two sites located in the Tumbes and Piura regions
of northern Peru with high predicted suitability for Cebus albifrons aequatorialis, the
Tumbes Reserved Zone (Fig. 1, locality 43) and the Cerros de Amotape National
Park, both of which are nationally protected areas within the North-West Peru
Biosphere Reserve (Fig. 5, Third Priority Region). Both sites are reportedly inhabited
by Cebus albifrons aequatorialis (Cornejo and de la Torre 2008; Encarnacion and
Cook 1998). The North-West Peru Biosphere contains the largest and most continuous tract of relatively undisturbed forest in the Tumbesian Region, and it should
therefore be considered of high conservation importance. Owing to its size, this
region could potentially harbor relatively large primate populations. At present, the
North-West Peru Biosphere apparently experiences lower levels of human disturbance than the other regions discussed here owing to its remoteness and the low
human population density in surrounding areas (Best and Kessler 1995), making it a
promising area for targeted conservation initiatives.
The fourth priority region is located in southern Ecuador, where there are numerous small areas of humid forest along the foothills of the Andes in the provinces of
Azuay, El Oro, and Guayas (Fig. 5, Fourth Priority Region). The largest of these
patches encompasses the 2500-ha private reserve Cerro de Hayas-Naranjal (Fig. 1,
locality 27), which is the only protected site in this region known to be inhabited by
Cebus albifrons aequatorialis (Albuja and Arcos 2007). The two other inhabited sites
in this region —Cerro Azul (Fig. 1, locality 34) and Cooperativa 31 de Agosto
(Fig. 1, locality 30)— do not currently have protected status. There are a number
of small, mostly unprotected forest fragments and private reserves in this region (Best
and Kessler 1995), although Cebus albifrons aequatorialis is apparently absent from
several that our model predicted as suitable (Albuja and Arcos 2007). We emphasize
that in addition to protecting inhabited core areas, an effective conservation plan for
Cebus albifrons aequatorialis should promote connectivity among the remaining
forest patches in these regions. Therefore, it is important not to ignore areas of lower
predicted suitability that may function as corridors between or buffer zones around
the core areas.
Acknowledgments F. A. Campos was supported by Alberta Innovates–Technology Futures and the
University of Calgary. This research was made possible through the following research grants awarded
to K. M. Jack: National Geographic Society Conservation Trust; Margot Marsh Biodiversity Foundation;
University Research Council at Appalachian State University; and the Committee on Research, George
Lurcy Fund, and the Stone Center for Latin American Studies at Tulane University. We thank Dr. Luis
Albuja and Rodrigo Acros for sharing locality data. We are grateful to the landowners and reserve
personnel that allowed us to carry out field work, with special thanks to Eric Von Horstman, Eudaldo
Loor, and the Cevallos-Martinez family. We also thank Zdanna King, Andrew Childers, Dale Morris, Sasha
Gilmore, Robert Lee, Kate Laakso, Marcelo Luque, Patricia Intriago Alvaro, Vincente Salvo, Melina
Costantino, Rafael Ángel, Alfonso Arguero Santos, and numerous field school students from Appalachian
Distribution Model and Conservation Plan for Ecuadorian Capuchin
State University for logistical support. We also thank our anonymous reviewers for numerous helpful
comments that improved the manuscript.
References
Albuja, V. L. (2002). Mamíferos del Ecuador. In G. Ceballos & J. A. Simonetti (Eds.), Diversidad y Conservación
de los Mamiferos Neotropicales (pp. 271–327). Mexico City: CONABIO-UNAM.
Albuja, V. L., & Arcos, D. R. (2007). Evaluación de las poblaciones de Cebus albifrons cf. aequatorialis en
los bosques suroccidentales Ecuatorianos. Politécnica, 27(4) Biología 7, 58–67.
Allen, J. A. (1914). New South American monkeys. Bulletin of the American Museum of Natural History,
33, 647–655.
Bartholome, E., & Belward, A. S. (2005). GLC2000: A new approach to global land cover mapping from
Earth observation data. International Journal of Remote Sensing, 26(9), 1959–1977.
Best, B. J., & Kessler, M. (1995). Biodiversity and conservation in Tumbesian Ecuador and Peru.
Cambridge, U.K.: BirdLife International.
Bierregaard, R. O. (2001). Lessons from Amazonia: The ecology and conservation of a fragmented forest.
New Haven, CT: Yale University Press.
Brooks, T. M., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., Rylands, A. B., Konstant, W.
R., et al. (2002). Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology,
16(4), 909–923.
Caro, T. M., & O'Doherty, G. (1999). On the use of surrogate species in conservation biology. Conservation
Biology, 13(4), 805–814.
Charlat, S., Thatcher, O. R., Hartmann, N., Patel, Y. G., Saillan, M., & Vooren, E. (2000). Survey of Alouatta
palliata at the Bilsa Biological Reserve, north-west Ecuador. Neotropical Primates, 8(1), 40–44.
Cornejo, F., & de la Torre, S. (2008). Cebus albifrons ssp. aequatorialis. Retrieved from www.iucnredlist.org
(Accessed February 28, 2013).
Dodson, C. H., & Gentry, A. H. (1991). Biological extinction in western Ecuador. Annals of the Missouri
Botanical Garden, 78(2), 273–295.
Elith, J., Graham, C. H., Anderson, R. P., Dudik, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography, 29(2), 129–151.
Encarnacion, F., & Cook, A. G. (1998). Primates of the tropical forest of the Pacific coast of Peru: The
Tumbes Reserved Zone. Primate Conservation, 18, 15–20.
Eva, H. D., de Miranda, E. E., Di Bella, C. M., Gond, V., Huber, O., Sgrenzaroli, M., et al. (2002). A
vegetation map of South America. Luxembourg: EUR 20159 EN, European Commission.
Fritz, S., Bartholomé, E., Belward, A., Hartley, A., Stibig, H. J., Eva, H., et al. (2003). Harmonisation,
mosaicing and production of the Global Land Cover 2000 database (beta version). European
Commission – Joint Research Centre.
Game, E. T., Kareiva, P., & Possingham, H. P. (2013). Six common mistakes in conservation priority
setting. Conservation Biology, 27(3), 480–485.
Gavilanez-Endara, M. M. (2006). Demografía, actividad y preferencia de hábitat de tres especies de
primates (Alouatta palliata, Ateles fusciceps y Cebus capucinus) en un bosque nublado del
Noroccidente Ecuatoriano. Pontifica Universidad Católica del Ecuador, Quito, Ecuador.
GLC (2003). Global Land Cover 2000 database. European Commission, Joint Research Centre, 2003.
Retrieved from http://bioval.jrc.ec.europa.eu/products/glc2000/glc2000.php
Hansen, M., DeFries, R., Townshend, J. R., Carroll, M., Dimiceli, C., & Sohlberg, R. (2007). Vegetation
continuous fields MOD44B, 2001 percent tree cover, Collection 4. College Park, Maryland: University
of Maryland.
Hansen, M. C., Defries, R. S., Townshend, J. R. G., & Sohlberg, R. (2000). Global land cover classification
at 1km spatial resolution using a classification tree approach. International Journal of Remote Sensing,
21(6–7), 1331–1364.
Harris, J. B. C., Tirira, D. G., Álvarez L, P. J., & Mendoza I, V. (2008). Altitudinal Range Extension for
Cebus albifrons (Primates: Cebidae) in Southern Ecuador. Neotropical Primates, 15(1), 22–24.
Hernandez, P. A., Graham, C. H., Master, L. L., & Albert, D. L. (2006). The effect of sample size and
species characteristics on performance of different species distribution modeling methods. Ecography,
29(5), 773–785.
F.A. Campos, K.M. Jack
Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G., & Jarvis, A. (2005). Very high resolution interpolated
climate surfaces for global land areas. International Journal of Climatology, 25(15), 1965–1978.
Hores, R. M. (2006). Census of nonhuman primate populations of Comuna El Pital, in south central
Ecuador (Cebus albifrons, Alouatta palliata). MA Thesis, Florida Atlantic University
Jack, K. M., & Campos, F. A. (2012). Distribution, abundance, and spatial ecology of the critically endangered
Ecuadorian capuchin (Cebus albifrons aequatorialis). Tropical Conservation Science, 5(2), 173–191.
Jimenez-Valverde, A., & Lobo, J. M. (2007). Threshold criteria for conversion of probability of species presence
to either-or presence-absence. Acta Oecologica-International Journal of Ecology, 31(3), 361–369.
Kleiman, D. G., & Mallinson, J. J. C. (1998). Recovery and management committees for lion tamarins:
Partnerships in conservation planning and implementation. Conservation Biology, 12(1), 27–38.
Lambeck, R. J. (1997). Focal species: A multi-species umbrella for nature conservation. Conservation
Biology, 11(4), 849–856.
Liknes, G. C., Perry, C. H., & Meneguzzo, D. M. (2010). Assessing tree cover in agricultural landscapes
using high-resolution aerial imagery. Journal of Terrestrial Observation, 2(1), Article 5.
Lindenmayer, D. B., Manning, A. D., Smith, P. L., Possingham, H. P., Fischer, J., Oliver, I., & McCarthy,
M. A. (2002). The focal-species approach and landscape restoration: A critique. Conservation Biology,
16(2), 338–345.
Liu, C. R., Berry, P. M., Dawson, T. P., & Pearson, R. G. (2005). Selecting thresholds of occurrence in the
prediction of species distributions. Ecography, 28(3), 385–393.
Loiselle, B. A., Howell, C. A., Graham, C. H., Goerck, J. M., Brooks, T., Smith, K. G., & Williams, P. H.
(2003). Avoiding pitfalls of using species distribution models in conservation planning. Conservation
Biology, 17(6), 1591–1600.
Manel, S., Williams, H. C., & Ormerod, S. J. (2001). Evaluating presence-absence models in ecology: The
need to account for prevalence. Journal of Applied Ecology, 38(5), 921–931.
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000). Biodiversity
hotspots for conservation priorities. Nature, 403(6772), 853–858.
Myneni, R. B., Hall, F. G., Sellers, P. J., & Marshak, A. L. (1995). The interpretation of spectral vegetation
indexes. Ieee Transactions on Geoscience and Remote Sensing, 33(2), 481–486.
Parker, T. A., & Carr, J. L., Eds. (1992). Status of forest remnants in the Cordillera de la Costa and
adjacent areas of southwestern Ecuador: Conservation International, RAP Working Papers 2.
Pearson, R. G., Raxworthy, C. J., Nakamura, M., & Peterson, A. T. (2007). Predicting species distributions
from small numbers of occurrence records: A test case using cryptic geckos in Madagascar. Journal of
Biogeography, 34(1), 102–117.
Peres, C. A. (2001). Synergistic effects of subsistence hunting and habitat fragmentation on Amazonian
forest vertebrates. Conservation Biology, 15(6), 1490–1505.
Phillips, S. J., Anderson, R. P., & Schapire, R. E. (2006). Maximum entropy modeling of species
geographic distributions. Ecological Modelling, 190(3–4), 231–259.
Portillo-Quintero, C. A., & Sánchez-Azofeifa, G. A. (2010). Extent and conservation of tropical dry forests
in the Americas. Biological Conservation, 143(1), 144–155.
Robinson, T. P., van Klinken, R. D., & Metternicht, G. (2010). Comparison of alternative strategies for
invasive species distribution modeling. Ecological Modelling, 221(19), 2261–2269.
Tirira, D. (Ed.). (2011). Libro Rojo de los mamíferos del Ecuador (2nd ed.). Quito: Fundación Mamíferos y
Conservación. Pontificia Universidad Católica del Ecuador & Ministerio del Ambiente del Ecuador.
Vidal-García, F., & Serio-Silva, J. (2011). Potential distribution of Mexican primates: Modeling the
ecological niche with the maximum entropy algorithm. Primates, 52(3), 261–270.
Wilson, K. A., Carwardine, J., & Possingham, H. P. (2009). Setting conservation priorities. Annals of the
New York Academy of Sciences, 1162(1), 237–264.
Wilson, K. A., McBride, M. F., Bode, M., & Possingham, H. P. (2006). Prioritizing global conservation
efforts. Nature, 440(7082), 337–340.